15 Jan, 2016

Ideas / Things to Investigate

  1. Updating Bayesian regression
  2. Quick way to get posterior means
  3. Indian Buffet Process (IBP) in relation to Non-negative Matrix Factorization (NMF)
  4. BNP Logistic Regression, modeling joint density, function regression?
  5. Spatial stats bayes + freq
  6. NLP
    • classifying authors in bible, with chapters as the documents
    • Naive Bayes: $P(c = c_j x_1,…,x_n) = \prod_{i=1}^n P(x_i c = c_j) P(c = c_j)$
    • Word Clouds
    • Stop Words

Notes:

  • From a practical perspective, Bayesian methods are used only when uncertainty is of interest. If the practitioner is only interested point estimation, frequentist methods are preferred.